Fig. 3

Performance mapping and selection of movements. a. The workspace for mapping motor performance is defined by all possible targets angles ϑ (targets 5 cm away from the workspace centre; blue dashed line) and all the possible start angles σ from each target (red dashed line, indicating the possible start locations relative to the target shown as a blue disc). Note that start locations define movement directions. b. An example of a 32 × 32 ϑ-by-σ performance map (interpolated and smoothed from the 8 × 8 performance metric PM3, indicated as black dots). Reddish colours indicate good performance, bluish colours poor performance. Note that opposite edges of the map are in fact contiguous, due to the circularity of angular data (0^o  = 360^o). The small white square, bottom left of map, indicates the coordinates of the start-to-target movement example shown in panel a (ϑ = 135, σ = 45; red Arrow). c. Example of performance-based selection of practiced movements (‘+’) according to the steepest gradients principle. The example selection shown was based on the PM3 map from a representative patient (also shown in panel b). Selected movements are located at regions where performance changes rapidly from higher to lower levels (relative to the participant’s overall performance). The graph across the top depicts the performance gradient measured at target direction of 202.5^o; as indicated by the white line on the map) and the corresponding selected movements (+). Note that selection is based on the vector sum of the local gradients across both map dimensions. Note also that the training sets for the study consisted of weighted selections from the PM2 and PM3 maps (see text and [16] for more details). d. The coordinates of centre-out movements which were selected for the control group (‘+’)